On Sat, Jan 3, 2015 at 2:36 PM, Pavel Machek <pavel@xxxxxx> wrote: > >> >> No. This is not what this patch does. >> >> >> >>> But changing glibc to do sleep(30); abort(); instead of abort(); to >> >>> slow down bruteforcing of canaries makes some kind of sense... and >> >>> should be ok by default. >> >> >> >> As I saidn only focusing one the specific stack canary case is not enough. >> > >> > Ok, so I am now saying "adding random delays to the kernel, hoping >> > they slow attacker down" is bad idea. Feel free to add my NAK to the >> > patch. >> >> The patch does not add random delays nor is hope involved. >> >> It has a very clear purpose, it makes brute force attacks to forking >> services unattractive. >> Exploits often use the fact that after fork() the child has the same memory >> as the parent and therefore an attacker can start fruitful brute force attacks >> to brute stack canaries, offsets, etc. as the new child will always have mostly >> the same memory layout as before. >> >> But I'll happily add your NAK to this series. > > Please do. > >> > If really neccessary, "kill_me_slowly()" syscall would be acceptable, >> > but it seems just sleep(); abort(); combination is enough. >> >> The goal of the patch is not to protect only against brute forcing the stack canary. >> It should protect against all kind of brute forcing using forking services. >> >> > glibc should cover 99% cases where this matters, please just fix glibc, >> > others will follow. >> >> There are a lot of systems out there without glibc. > > Only "interesting" systems that are without glibc are androids, and > they usually run very old kernels. > > If you implement sleep() in glibc, distros will enable it and you'll > protect all the desktop users. As an attempt to help end this particular line of debate: putting the sleep in glibc won't work. The point isn't to make the crashed process crash more slowly; it's to limit the rate at which *new* siblings can be forked and crashed as a canary or ASLR brute-force probe. IOW, adding a sleep call to glibc slows down the wrong thing. Also, trying to get libc to take action on a plain old segfault is a giant mess, because it involves mucking with signal handling, which glibc really has no business doing by default. Also, this patch is missing a bit, I think. We really want to control the total rate of crashes. This patch imposes a delay per crash, but AFAICS it would still be possible for an attacker to coerce a forking server to fork, say, 10k children, then probe all of them, then wait 30 seconds and repeat. --Andy -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
